Found 12 posts.
Search results Show results as topic list.
Technical Discussion » custom crowd trigger node
- Karl Richter
- 12 posts
- Offline
While it is true I could just use the VEX mode, I see in the manual that there is a way to make your own. So now I want to! A common frustration with the Houdini manual, I'm sure.
Edited by Karl Richter - July 12, 2023 12:21:58
Technical Discussion » custom crowd trigger node
- Karl Richter
- 12 posts
- Offline
I'm interested in creating my own custom crowd trigger node as the current node is somewhat limited in options. I know I can use the VEX mode to get more detailed, but I was hoping to setup something like the pop vop but for setting triggers. In the manual it says: "A user-defined asset can also be used as a trigger if it has a triggerattrib parameter specifying the name of the point attribute that contains its trigger value." But when I setup this parameter, it still doesn't seem to work, with the error "cant find trigger name" Any ideas?
Thanks, Karl
Thanks, Karl
Technical Discussion » Houdini 18 really slow to launch with Arnold
- Karl Richter
- 12 posts
- Offline
Houdini has always been a little slow to launch for us, about 1.5 minutes, but I kind of attributed it (with no real evidence) to the large python code base on the network that is added to the PATH environment variable. With 18, launch times are around 3 minutes. These times are with the Arnold htoa plugin installed. If the Arnold plugin is excluded, launch times are much faster.
So my question is, is there any way to get some kind of debug info on what may be causing these long startups? I'm impatient and think of all the stuff I could get done in those 3 minutes!
Thanks,
Karl
So my question is, is there any way to get some kind of debug info on what may be causing these long startups? I'm impatient and think of all the stuff I could get done in those 3 minutes!
Thanks,
Karl
Technical Discussion » Odd tumbling behaviour
- Karl Richter
- 12 posts
- Offline
Thanks for the tip Pabcou, that was totally it.
The problem seems to be a weird interaction between the current version of Houdini and an older Redshift plugin. The fix is to install the version of Redshift that is specifically built against 17.5.293.
The problem seems to be a weird interaction between the current version of Houdini and an older Redshift plugin. The fix is to install the version of Redshift that is specifically built against 17.5.293.
Technical Discussion » Odd tumbling behaviour
- Karl Richter
- 12 posts
- Offline
Yes. I have tried all the different “Tumbling Methods” and “View State Operation Defaults” in the preferences, but it makes no difference. I still have my previous houdini build installed and its fine.
Technical Discussion » Odd tumbling behaviour
- Karl Richter
- 12 posts
- Offline
After updating from 7.5.258 to 7.5.293 I have noticed this weird behaviour that occurs when you tumble the camera after zooming in. The camera seems to pop to a slightly different location and a bunch of roll is added in, as far as I can tell, a totally random way. It's really driving me nuts and i'm surprised that no one else has noticed it (at least on the forums I go to).
here is a gif illustrating the issue:
It's weird that it only happens after a camera zoom. You can pan and tumble all you want, but after a zoom, the next tumble will pop the camera into a weird spot/rotation. Is anyone else having this problem?
Thanks,
Karl
here is a gif illustrating the issue:
It's weird that it only happens after a camera zoom. You can pan and tumble all you want, but after a zoom, the next tumble will pop the camera into a weird spot/rotation. Is anyone else having this problem?
Thanks,
Karl
Edited by Karl Richter - July 24, 2019 18:50:08
Technical Discussion » Vellum collision penetration
- Karl Richter
- 12 posts
- Offline
Vellum collision detection is overall very good, except when a pesky animator does something silly like move an arm through the torso or some other kind of interpenetration (the camera can't see it so it's ok, right?). Once a bit of cloth gets stuck inside the collision geo, it can be very difficult to get out. If it does get out, sometimes the cloth will act strangely in that area. I made a VDB based network setup that tries to push the cloth back out that works OK in most cases, but it's slow and kind of a hack.
Other cloth systems I have used (nCloth, Syflex) seem to handle this kind of issue a little better.
What do you folks do when your vellum cloth gets stuck in some geo (just making sure i'm not missing something)?
Thanks,
Karl
Other cloth systems I have used (nCloth, Syflex) seem to handle this kind of issue a little better.
What do you folks do when your vellum cloth gets stuck in some geo (just making sure i'm not missing something)?
Thanks,
Karl
Technical Discussion » Vellum - Match Animation & Gravity
- Karl Richter
- 12 posts
- Offline
Your problem is that the animation you are trying to match is floating above your sim. It looks like you want to pull the sheets after they have settled on the collision object. In real life, you would grab the cloth by the edge and pull. In the attached example, I setup a very simple constraint that turns on at frame 75 and tries pulls all your sheets off.
for the initial frame, you could run the sim up to the point you want and cache out that geo, and then use that as the starting input cloth shape. To have it accurately preserve your cloth starting rest constraints, check out the “vellum rest blend” node.
Karl
for the initial frame, you could run the sim up to the point you want and cache out that geo, and then use that as the starting input cloth shape. To have it accurately preserve your cloth starting rest constraints, check out the “vellum rest blend” node.
Karl
Edited by Karl Richter - May 6, 2019 14:31:21
Technical Discussion » vellum work flow
- Karl Richter
- 12 posts
- Offline
We normally use nCloth for our cloth sims, but I thought I would give Vellum a try for an upcoming project. I am getting really good results, but there seems to be a lack of tools to transfer the cloth simulation from the proxy simulated cloth to the rendered asset.
The way things are currently setup, it seems like it is expected that you would render a subdivided version of your simulated cloth. This will not work for us as the modeled garment has thickness and seams and other details modeled into it that are not simulation friendly.
With nCloth, I would use a wrap deformer to drive the rendered cloth. There is a “cloth deform” node in Houdini, but it seems old and I couldn't find anyway to adjust the weights. Ultimately, I get better results using the point deformer, but I still feel like there should be something better…
So, what is the prescribed way to drive your hires cloth from your proxy sim?
Thanks,
Karl
The way things are currently setup, it seems like it is expected that you would render a subdivided version of your simulated cloth. This will not work for us as the modeled garment has thickness and seams and other details modeled into it that are not simulation friendly.
With nCloth, I would use a wrap deformer to drive the rendered cloth. There is a “cloth deform” node in Houdini, but it seems old and I couldn't find anyway to adjust the weights. Ultimately, I get better results using the point deformer, but I still feel like there should be something better…
So, what is the prescribed way to drive your hires cloth from your proxy sim?
Thanks,
Karl
Technical Discussion » Flip Solver Surface Tension Blur
- Karl Richter
- 12 posts
- Offline
It seems like you want to be able to control the amount of “surface tension” in different areas of your flip sim. I recently had a situation where I wanted to do the same thing. I didn't find a way to directly control tension, but I did find a quick and dirty work around, and that is using the “density” particle attribute. if you lower or increase the particle density attribute, it affects how much surface tension affects your sim, but may have other side effect.
Karl
Karl
Technical Discussion » Particles follow volume
- Karl Richter
- 12 posts
- Offline
Well, I couldn't leave it at that. Here is a more optimized version with reseeding. The main issue with this method is that when disconnected areas of the point cloud appear and disappear, the tracking breaks down. This is not a perfect solution, but I think you could use it to generate a decent velocity field….
Technical Discussion » Particles follow volume
- Karl Richter
- 12 posts
- Offline
What an interesting problem. Here is my quick stab at it. Could be improved with a reseeding routine, but it's a good start.
Karl
Karl
-
- Quick Links